change default integration method to rk45

for DiffusionModel and FlowMatching. Euler shows significant deviations
when computing the log-prob, which risks misleading users regarding the
performance of the networks.

rk45 is slower, but the problem is heavily reduced with this method.

Author: vpratz

bayesflow/networks/diffusion_model/diffusion_model.py

@@ -44,7 +44,7 @@ class DiffusionModel(InferenceNetwork):
     }
 
     INTEGRATE_DEFAULT_CONFIG = {
-        "method": "euler",
+        "method": "rk45",
         "steps": 100,
     }
 

bayesflow/networks/flow_matching/flow_matching.py

@@ -53,7 +53,7 @@ class FlowMatching(InferenceNetwork):
     }
 
     INTEGRATE_DEFAULT_CONFIG = {
-        "method": "euler",
+        "method": "rk45",
         "steps": 100,
     }